The protection of instream flows, the flow of water in natural river channels, is a controversial environmental issue throughout the US West. This is especially the case in New Mexico, which is unique in that no legal avenue for protection exists. This contingent valuation study investigates the non-market benefits of protecting minimum instream flows in New Mexico; the elicited values are expected to have a significant non-use component. The original dichotomous choice contingent valuation telephone survey was conducted in February 1995, and used a voluntary contribution trust fund format. Using the same 2×2 experimental design, computer-assisted telephone interview system and sampling frame of New Mexico residents, the survey was then replicated in February 1996. As a measure of theoretical validity, we test the sensitivity in valuation responses to a change in the scope of the good (nested geographical components). Split-sample statistical tests of scope remain relatively rare in contingent valuation studies involving nonuse values, especially when combined with other tests of validity and reliability using the same sampling frame and survey instrument. Thus, corollary tests of sensitivity in valuation responses to information about the collective nature of providing the good (a group-size reminder) and the temporal reliability of results are also conducted. Despite the explicit ‘contributions frame’, there is no evidence to support the simple contribution model for interpreting valuation responses.